Hittite - good products, horrible documentation?

[daqq]

Hi guys,

Is it just me, or does Hittite (bought by Analog Devices) seriously lack in the documentation department? I'm looking at some of their stuff and, well, the datasheets are vague and incomplete. See:

http://www.analog.com/media/en/technical-documentation/data-sheets/hmc470A.pdf

It's seven pages, one of which is an automatically inserted Analog Devices page full of links and another one is the evaluation board layout. The package drawing is a crudely inserted JPG that suffered due to poor compression. There are helpful hints like:

External capacitor to ground is required. Select value for lowest frequency of operation. Place capacitor as close to pins as possible.

For 60MHz is this 1nF? 1pF? 330pF? 10TF? Is there any formula to calculate this? Or a table? Or something?

This pin is DC coupled and matched to 50 Ohm. Blocking capacitors are required. Select value based on lowest frequency of operation.

So, is it a DC or AC coupled pin? I'm confused, if it requires a DC blocking capacitor, then the attenuator does not go from DC, but from some frequency.

These pins should be connected to PCB RF ground to maximize performance.

Then don't name them N/C?

There are no timing diagrams, examples of waveforms (I'd love to see if there are any glitches during attenuation changes in the output signal) etc. Basically, the datasheet(s) assume that the user feels safe assuming a lot of stuff or does not care about a lot of stuff.

A different device:

For  frequencies  less  than  700  MHz,  the  use  of  ACG  capacitors  C4,  C5  and  C6  is  recommended.  For  frequencies greater than 700 MHz, the HMC792ALP4E has similar performance with and without the ACG capacitors

What's "similar"? Is it 0.42 dB different with a peak at 712MHz? Does it spawn demons from pin 4 but otherwise works exactly the same?

Basically it seems to me that the datasheets from Hitite are abridged versions really.

/rant

By the way, does anyone know of a small switchable attenuator ranging from 0-15dB attenuation, preferably with no external components? It needs to work at least in the 55-65 MHz range, beyond that I don't care, 50 Ohms IO, 2dB or smaller steps, parralel interface and preferably fast state changes without glitches.



David

[KE5FX]

Suggest looking at www.psemi.com - their data sheets aren't much better but their parts tend to be.  Hittite is largely still GaAs oriented in an age where CMOS is making inroads into RF/microwave.

[dmills]

Minicircuits have some switched attenuators that are happy in the 6M band, parallel or SPI variants are available, and they do work (You need the usual decoupling passives and a DC block cap in and out, should not be an issue at ~50MHz).

I assume a QFN is acceptable packaging.

Regards, Dan.

[1design]

I used probably over 100 Hittite products over the years and the datasheets always had all the required information in them. So does this one.
I never experienced any glitching while switching attenuators, they are just fets switching on and off. The capacitors are very basic RF calculations, do not belong in the datasheet as they are not part of the IC performance, but a design decision.

With a basic understanding of how packages are made it is clear that those pins are internally N/C but good practice is to ground them so no stray voltage voltage forms on them in this case.

The input and output pins are DC coupled so they need to be decoupled, again, standard practice with most components. It is matched to 50 ohms so a basic RC 1st order filter equation will give you the answer regarding capacitor size, but watch out for phase shifts if you are closer than a decade to the minimum frequency of interest.

Minicircuits also works very well and have used plenty of their components over the years, again, you will have to do some work to match components to what you need the part to do in your specific case.


BR

[ebclr]

With the acquisition of Hittite Microwave, the industry's broadest portfolio of RF and Microwave Solutions are now available through Analog Devices's worldwide

Let's wait, on the long run probably will have AD documentation quality on the survived products from  Hittite

[karoru]

At least new designs, for older ones they could just slap AD logo on the first page (in best case via real wordprocessing, in worst case using some old scanned copy) and call it a day

[eb4fbz]

I used probably over 100 Hittite products over the years and the datasheets always had all the required information in them. So does this one.
I never experienced any glitching while switching attenuators, they are just fets switching on and off. The capacitors are very basic RF calculations, do not belong in the datasheet as they are not part of the IC performance, but a design decision.

With a basic understanding of how packages are made it is clear that those pins are internally N/C but good practice is to ground them so no stray voltage voltage forms on them in this case.

The input and output pins are DC coupled so they need to be decoupled, again, standard practice with most components. It is matched to 50 ohms so a basic RC 1st order filter equation will give you the answer regarding capacitor size, but watch out for phase shifts if you are closer than a decade to the minimum frequency of interest.

Minicircuits also works very well and have used plenty of their components over the years, again, you will have to do some work to match components to what you need the part to do in your specific case.


BR

Same here, dozen of RF designs using Hittite ICs (att, lna, power amps, mixers, IQ mod/demods...) and never had a problem, their datasheets are fine and it's nice to have the I/Os equivalent circuit. Never had to contact their support department.

[radar_macgyver]

Their PLLs can be a pain to program. If you go by the datasheet and come up with values to program into the registers, things just don't work. The right way, apparently, is to use the eval board software, which generates a table of values that you write to the registers. Comparing these to the ones I had calculated from the data sheet reveals that bits marked "not used" in the data sheet had been flipped, and that was the only difference. Had this happen with an HMC834.

[randsl]

Usually Hittite chips works well without much effort "just out of the box".

So far I've used about a dozen of those.
I've had only one oscillation problem with one of the Ka band LNA due to lack of decoupling caps. I used the values used in the eval. board, but didn't work. Had to add few more to stop the oscillation. May be there was a contribution from the layout as well.

However, for a novice designer, it may give the feeling of lack of documentation. But I think the real situation is the datasheet has all the information.